RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPO

TL;DR

RAVEN introduces a novel training framework and reinforcement learning method for real-time autoregressive video extrapolation, significantly improving long-horizon generation quality.

Contribution

The paper proposes RAVEN, a training-time test framework for better history representation, and CM-GRPO, a reinforcement learning approach that enhances video extrapolation performance.

Findings

RAVEN outperforms recent causal video distillation baselines in quality and semantic evaluations.

CM-GRPO provides additional gains when combined with RAVEN.

Experiments show improved long-horizon video generation quality.

Abstract

Causal autoregressive video diffusion models support real-time streaming generation by extrapolating future chunks from previously generated content. Distilling such generators from high-fidelity bidirectional teachers yields competitive few-step models, yet a persistent gap between the history distributions encountered during training and those arising at inference constrains generation quality over long horizons. We introduce the Real-time Autoregressive Video Extrapolation Network (RAVEN), a training-time test framework that repacks each self rollout into an interleaved sequence of clean historical endpoints and noisy denoising states. This formulation aligns training attention with inference-time extrapolation and allows downstream chunk losses to supervise the history representations on which future predictions depend. We further propose Consistency-model Group Relative Policy Optimization (CM-GRPO), which reformulates a consistency sampling step as a conditional Gaussian transition and applies online Reinforcement Learning (RL) directly to this kernel, avoiding the Euler-Maruyama auxiliary process adopted in prior flow-model RL formulations. Experiments demonstrate that RAVEN surpasses recent causal video distillation baselines across quality, semantic, and dynamic degree evaluations, and that CM-GRPO provides further gains when combined with RAVEN.